Transfer learning using Mask R-CNN on Custom Coco Dataset.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import torch\n",
    "import torch.utils.data\n",
    "import torchvision\n",
    "from PIL import Image, ImageDraw\n",
    "from pycocotools.coco import COCO\n",
    "import numpy as np\n",
    "\n",
    "class CocoDataset(torch.utils.data.Dataset):\n",
    "    def __init__(self, root, annotation, transforms=None):\n",
    "        self.root = root\n",
    "        self.transforms = transforms\n",
    "        self.coco = COCO(annotation)\n",
    "        self.ids = list(sorted(self.coco.imgs.keys()))\n",
    "\n",
    "    def __getitem__(self, index):\n",
    "        coco = self.coco\n",
    "        # image id\n",
    "        img_id = self.ids[index]\n",
    "        # fetch all annotations belonging to an image id\n",
    "        ann_ids = coco.getAnnIds(imgIds=img_id)\n",
    "        # coco annotations\n",
    "        annotations = coco.loadAnns(ann_ids)\n",
    "        # path of input image\n",
    "        path = coco.loadImgs(img_id)[0]['file_name']\n",
    "        # open the input image\n",
    "        img = Image.open(os.path.join(self.root, path))\n",
    "        img_width, img_height = img.size\n",
    "\n",
    "        # number of objects in the image\n",
    "        num_objs = len(annotations)\n",
    "        instance_masks = []\n",
    "        for annotation in annotations:\n",
    "            # this is our canvas\n",
    "            mask = Image.new('1', (img_width, img_height))\n",
    "            mask_draw = ImageDraw.Draw(mask, '1')\n",
    "            for segmentation in annotation['segmentation']:\n",
    "                mask_draw.polygon(segmentation, fill=1)\n",
    "                bool_array = np.array(mask) > 0\n",
    "                instance_masks.append(bool_array)\n",
    "        # a set of binary masks (one per object)\n",
    "        masks = torch.as_tensor(instance_masks, dtype=torch.uint8)\n",
    "        \n",
    "        # bounding boxes\n",
    "        boxes = []\n",
    "        for i in range(num_objs):\n",
    "            xmin = annotations[i]['bbox'][0]\n",
    "            ymin = annotations[i]['bbox'][1]\n",
    "            xmax = xmin + annotations[i]['bbox'][2]\n",
    "            ymax = ymin + annotations[i]['bbox'][3]\n",
    "            boxes.append([xmin, ymin, xmax, ymax])\n",
    "        boxes = torch.as_tensor(boxes, dtype=torch.float32)\n",
    "        labels = torch.ones((num_objs,), dtype=torch.int64)\n",
    "        img_id = torch.tensor([img_id])\n",
    "        areas = []\n",
    "        for i in range(num_objs):\n",
    "            areas.append(annotations[i]['area'])\n",
    "        areas = torch.as_tensor(areas, dtype=torch.float32)\n",
    "        iscrowd = torch.zeros((num_objs,), dtype=torch.int64)\n",
    "        \n",
    "        \n",
    "        # target in dictionary format\n",
    "        target = {}\n",
    "        target[\"boxes\"] = boxes\n",
    "        target[\"labels\"] = labels\n",
    "        target[\"masks\"] = masks\n",
    "        target[\"image_id\"] = img_id\n",
    "        target[\"area\"] = areas\n",
    "        target[\"iscrowd\"] = iscrowd\n",
    "\n",
    "        if self.transforms is not None:\n",
    "            img, target = self.transforms(img, target)\n",
    "\n",
    "        return img, target\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.ids)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Download TorchVision repo to use some files from\n",
    "# references/detection\n",
    "#git clone https://github.com/pytorch/vision.git\n",
    "#cd vision\n",
    "#git checkout v0.3.0\n",
    "\n",
    "#cp references/detection/utils.py ../\n",
    "#cp references/detection/transforms.py ../\n",
    "#cp references/detection/coco_eval.py ../\n",
    "#cp references/detection/engine.py ../\n",
    "#cp references/detection/coco_utils.py ../\n",
    "\n",
    "#rm vision"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from engine import train_one_epoch, evaluate\n",
    "import utils\n",
    "import transforms as T\n",
    "\n",
    "# In my case, just added ToTensor\n",
    "def get_transform():\n",
    "    transforms = []\n",
    "    # converts the image, a PIL image, into a PyTorch Tensor\n",
    "    transforms.append(T.ToTensor())\n",
    "    return T.Compose(transforms)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load data "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data_dir = \"<your custom images go here>\"\n",
    "train_coco = \"<your custom coco annotations go here>\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loading annotations into memory...\n",
      "Done (t=0.00s)\n",
      "creating index...\n",
      "index created!\n"
     ]
    }
   ],
   "source": [
    "# create own Dataset\n",
    "full_dataset = CocoDataset(root=train_data_dir,\n",
    "                          annotation=train_coco,\n",
    "                          transforms=get_transform())\n",
    "\n",
    "train_size = int(0.8 * len(full_dataset))\n",
    "test_size = len(full_dataset) - train_size\n",
    "dataset, dataset_test = torch.utils.data.random_split(full_dataset, [train_size, test_size])\n",
    "\n",
    "\n",
    "# define training and validation data loaders\n",
    "data_loader = torch.utils.data.DataLoader(\n",
    "    dataset, batch_size=2, shuffle=True, num_workers=4,\n",
    "    collate_fn=utils.collate_fn)\n",
    "\n",
    "data_loader_test = torch.utils.data.DataLoader(\n",
    "    dataset_test, batch_size=1, shuffle=False, num_workers=4,\n",
    "    collate_fn=utils.collate_fn)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Fine Tuning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torchvision\n",
    "from torchvision.models.detection.faster_rcnn import FastRCNNPredictor\n",
    "from torchvision.models.detection.mask_rcnn import MaskRCNNPredictor\n",
    "\n",
    "      \n",
    "def get_instance_segmentation_model(num_classes):\n",
    "    # load an instance segmentation model pre-trained on COCO\n",
    "    model = torchvision.models.detection.maskrcnn_resnet50_fpn(pretrained=True)\n",
    "\n",
    "    # get the number of input features for the classifier\n",
    "    in_features = model.roi_heads.box_predictor.cls_score.in_features\n",
    "    # replace the pre-trained head with a new one\n",
    "    model.roi_heads.box_predictor = FastRCNNPredictor(in_features, num_classes)\n",
    "\n",
    "    # now get the number of input features for the mask classifier\n",
    "    in_features_mask = model.roi_heads.mask_predictor.conv5_mask.in_channels\n",
    "    hidden_layer = 256\n",
    "    # and replace the mask predictor with a new one\n",
    "    model.roi_heads.mask_predictor = MaskRCNNPredictor(in_features_mask,\n",
    "                                                       hidden_layer,\n",
    "                                                       num_classes)\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Now let's instantiate the model and the optimizer\n",
    "device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')\n",
    "\n",
    "\n",
    "# our dataset has two classes only - background and person\n",
    "num_classes = 2\n",
    "\n",
    "# get the model using our helper function\n",
    "model = get_instance_segmentation_model(num_classes)\n",
    "# move model to the right device\n",
    "model.to(device)\n",
    "\n",
    "# construct an optimizer\n",
    "params = [p for p in model.parameters() if p.requires_grad]\n",
    "optimizer = torch.optim.SGD(params, lr=0.005,\n",
    "                            momentum=0.9, weight_decay=0.0005)\n",
    "\n",
    "# and a learning rate scheduler which decreases the learning rate by\n",
    "# 10x every 3 epochs\n",
    "lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer,\n",
    "                                               step_size=3,\n",
    "                                               gamma=0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: [0]  [0/6]  eta: 0:00:05  lr: 0.001004  loss: 3.4752 (3.4752)  loss_classifier: 0.4713 (0.4713)  loss_box_reg: 0.0909 (0.0909)  loss_mask: 2.9056 (2.9056)  loss_objectness: 0.0051 (0.0051)  loss_rpn_box_reg: 0.0023 (0.0023)  time: 0.9062  data: 0.2868  max mem: 2138\n",
      "Epoch: [0]  [5/6]  eta: 0:00:00  lr: 0.005000  loss: 1.1839 (2.0219)  loss_classifier: 0.1406 (0.2677)  loss_box_reg: 0.0868 (0.0901)  loss_mask: 0.9293 (1.6493)  loss_objectness: 0.0079 (0.0098)  loss_rpn_box_reg: 0.0030 (0.0051)  time: 0.5695  data: 0.0499  max mem: 2413\n",
      "Epoch: [0] Total time: 0:00:03 (0.5741 s / it)\n",
      "creating index...\n",
      "index created!\n",
      "Test:  [0/4]  eta: 0:00:01  model_time: 0.1123 (0.1123)  evaluator_time: 0.0043 (0.0043)  time: 0.2685  data: 0.1508  max mem: 2413\n",
      "Test:  [3/4]  eta: 0:00:00  model_time: 0.1107 (0.1113)  evaluator_time: 0.0025 (0.0034)  time: 0.1543  data: 0.0389  max mem: 2413\n",
      "Test: Total time: 0:00:00 (0.1633 s / it)\n",
      "Averaged stats: model_time: 0.1107 (0.1113)  evaluator_time: 0.0025 (0.0034)\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "IoU metric: bbox\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.382\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.788\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.157\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.401\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.390\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.325\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.550\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.550\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.400\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.700\n",
      "IoU metric: segm\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.459\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.788\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.611\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.550\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.432\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.425\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.600\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.600\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.550\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.650\n",
      "Epoch: [1]  [0/6]  eta: 0:00:04  lr: 0.005000  loss: 0.5964 (0.5964)  loss_classifier: 0.1440 (0.1440)  loss_box_reg: 0.1322 (0.1322)  loss_mask: 0.3056 (0.3056)  loss_objectness: 0.0096 (0.0096)  loss_rpn_box_reg: 0.0049 (0.0049)  time: 0.7901  data: 0.2766  max mem: 2413\n",
      "Epoch: [1]  [5/6]  eta: 0:00:00  lr: 0.005000  loss: 0.4088 (0.4723)  loss_classifier: 0.0731 (0.0850)  loss_box_reg: 0.0755 (0.0763)  loss_mask: 0.2519 (0.2983)  loss_objectness: 0.0068 (0.0069)  loss_rpn_box_reg: 0.0044 (0.0058)  time: 0.6181  data: 0.0477  max mem: 2413\n",
      "Epoch: [1] Total time: 0:00:03 (0.6243 s / it)\n",
      "creating index...\n",
      "index created!\n",
      "Test:  [0/4]  eta: 0:00:01  model_time: 0.1367 (0.1367)  evaluator_time: 0.0043 (0.0043)  time: 0.3305  data: 0.1884  max mem: 2413\n",
      "Test:  [3/4]  eta: 0:00:00  model_time: 0.1272 (0.1346)  evaluator_time: 0.0043 (0.0057)  time: 0.1905  data: 0.0481  max mem: 2413\n",
      "Test: Total time: 0:00:00 (0.1976 s / it)\n",
      "Averaged stats: model_time: 0.1272 (0.1346)  evaluator_time: 0.0043 (0.0057)\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "IoU metric: bbox\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.242\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.829\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.043\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.451\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.187\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.225\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.450\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.450\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.450\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.450\n",
      "IoU metric: segm\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.427\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.829\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.419\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.495\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.374\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.400\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.575\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.575\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.600\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.550\n",
      "Epoch: [2]  [0/6]  eta: 0:00:05  lr: 0.005000  loss: 0.3136 (0.3136)  loss_classifier: 0.0435 (0.0435)  loss_box_reg: 0.0497 (0.0497)  loss_mask: 0.2068 (0.2068)  loss_objectness: 0.0108 (0.0108)  loss_rpn_box_reg: 0.0027 (0.0027)  time: 0.8896  data: 0.2482  max mem: 2413\n",
      "Epoch: [2]  [5/6]  eta: 0:00:00  lr: 0.005000  loss: 0.3604 (0.3696)  loss_classifier: 0.0516 (0.0534)  loss_box_reg: 0.0590 (0.0647)  loss_mask: 0.2068 (0.2383)  loss_objectness: 0.0071 (0.0072)  loss_rpn_box_reg: 0.0055 (0.0061)  time: 0.6124  data: 0.0440  max mem: 2413\n",
      "Epoch: [2] Total time: 0:00:03 (0.6182 s / it)\n",
      "creating index...\n",
      "index created!\n",
      "Test:  [0/4]  eta: 0:00:01  model_time: 0.1247 (0.1247)  evaluator_time: 0.0113 (0.0113)  time: 0.2897  data: 0.1527  max mem: 2413\n",
      "Test:  [3/4]  eta: 0:00:00  model_time: 0.1247 (0.1247)  evaluator_time: 0.0097 (0.0120)  time: 0.1761  data: 0.0385  max mem: 2413\n",
      "Test: Total time: 0:00:00 (0.1851 s / it)\n",
      "Averaged stats: model_time: 0.1247 (0.1247)  evaluator_time: 0.0097 (0.0120)\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "IoU metric: bbox\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.372\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.842\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.129\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.401\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.337\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.400\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.525\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.525\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.400\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.650\n",
      "IoU metric: segm\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.525\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.842\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.781\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.596\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.460\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.450\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.675\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.675\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.650\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.700\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: [3]  [0/6]  eta: 0:00:04  lr: 0.000500  loss: 0.3636 (0.3636)  loss_classifier: 0.0416 (0.0416)  loss_box_reg: 0.0684 (0.0684)  loss_mask: 0.2462 (0.2462)  loss_objectness: 0.0024 (0.0024)  loss_rpn_box_reg: 0.0050 (0.0050)  time: 0.7916  data: 0.2846  max mem: 2413\n",
      "Epoch: [3]  [5/6]  eta: 0:00:00  lr: 0.000500  loss: 0.2628 (0.2894)  loss_classifier: 0.0416 (0.0407)  loss_box_reg: 0.0589 (0.0623)  loss_mask: 0.1589 (0.1777)  loss_objectness: 0.0024 (0.0028)  loss_rpn_box_reg: 0.0050 (0.0059)  time: 0.5476  data: 0.0490  max mem: 2413\n",
      "Epoch: [3] Total time: 0:00:03 (0.5522 s / it)\n",
      "creating index...\n",
      "index created!\n",
      "Test:  [0/4]  eta: 0:00:01  model_time: 0.1274 (0.1274)  evaluator_time: 0.0132 (0.0132)  time: 0.2882  data: 0.1465  max mem: 2413\n",
      "Test:  [3/4]  eta: 0:00:00  model_time: 0.1191 (0.1232)  evaluator_time: 0.0100 (0.0128)  time: 0.1739  data: 0.0369  max mem: 2413\n",
      "Test: Total time: 0:00:00 (0.1808 s / it)\n",
      "Averaged stats: model_time: 0.1191 (0.1232)  evaluator_time: 0.0100 (0.0128)\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "IoU metric: bbox\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.376\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.851\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.129\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.401\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.345\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.400\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.525\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.525\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.400\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.650\n",
      "IoU metric: segm\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.529\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.851\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.752\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.592\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.480\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.450\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.675\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.675\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.650\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.700\n",
      "Epoch: [4]  [0/6]  eta: 0:00:04  lr: 0.000500  loss: 0.2479 (0.2479)  loss_classifier: 0.0381 (0.0381)  loss_box_reg: 0.0437 (0.0437)  loss_mask: 0.1591 (0.1591)  loss_objectness: 0.0038 (0.0038)  loss_rpn_box_reg: 0.0032 (0.0032)  time: 0.7363  data: 0.2374  max mem: 2413\n",
      "Epoch: [4]  [5/6]  eta: 0:00:00  lr: 0.000500  loss: 0.2576 (0.2800)  loss_classifier: 0.0381 (0.0386)  loss_box_reg: 0.0522 (0.0610)  loss_mask: 0.1591 (0.1714)  loss_objectness: 0.0036 (0.0033)  loss_rpn_box_reg: 0.0051 (0.0057)  time: 0.5400  data: 0.0422  max mem: 2413\n",
      "Epoch: [4] Total time: 0:00:03 (0.5448 s / it)\n",
      "creating index...\n",
      "index created!\n",
      "Test:  [0/4]  eta: 0:00:01  model_time: 0.1271 (0.1271)  evaluator_time: 0.0125 (0.0125)  time: 0.2892  data: 0.1485  max mem: 2413\n",
      "Test:  [3/4]  eta: 0:00:00  model_time: 0.1271 (0.1263)  evaluator_time: 0.0115 (0.0131)  time: 0.1786  data: 0.0383  max mem: 2413\n",
      "Test: Total time: 0:00:00 (0.1874 s / it)\n",
      "Averaged stats: model_time: 0.1271 (0.1263)  evaluator_time: 0.0115 (0.0131)\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "IoU metric: bbox\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.353\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.851\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.129\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.350\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.344\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.375\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.500\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.500\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.350\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.650\n",
      "IoU metric: segm\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.545\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.851\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.851\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.640\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.472\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.475\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.675\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.675\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.700\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.650\n",
      "Epoch: [5]  [0/6]  eta: 0:00:04  lr: 0.000500  loss: 0.2482 (0.2482)  loss_classifier: 0.0334 (0.0334)  loss_box_reg: 0.0690 (0.0690)  loss_mask: 0.1364 (0.1364)  loss_objectness: 0.0025 (0.0025)  loss_rpn_box_reg: 0.0068 (0.0068)  time: 0.7432  data: 0.2379  max mem: 2413\n",
      "Epoch: [5]  [5/6]  eta: 0:00:00  lr: 0.000500  loss: 0.2482 (0.2699)  loss_classifier: 0.0354 (0.0384)  loss_box_reg: 0.0563 (0.0625)  loss_mask: 0.1476 (0.1584)  loss_objectness: 0.0033 (0.0051)  loss_rpn_box_reg: 0.0033 (0.0055)  time: 0.5410  data: 0.0423  max mem: 2413\n",
      "Epoch: [5] Total time: 0:00:03 (0.5470 s / it)\n",
      "creating index...\n",
      "index created!\n",
      "Test:  [0/4]  eta: 0:00:01  model_time: 0.1294 (0.1294)  evaluator_time: 0.0157 (0.0157)  time: 0.2926  data: 0.1465  max mem: 2413\n",
      "Test:  [3/4]  eta: 0:00:00  model_time: 0.1218 (0.1263)  evaluator_time: 0.0134 (0.0141)  time: 0.1789  data: 0.0374  max mem: 2413\n",
      "Test: Total time: 0:00:00 (0.1873 s / it)\n",
      "Averaged stats: model_time: 0.1218 (0.1263)  evaluator_time: 0.0134 (0.0141)\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "IoU metric: bbox\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.340\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.862\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.086\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.350\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.320\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.375\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.500\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.500\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.350\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.650\n",
      "IoU metric: segm\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.563\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.862\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.862\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.650\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.502\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.475\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.675\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.675\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.700\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.650\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: [6]  [0/6]  eta: 0:00:04  lr: 0.000050  loss: 0.2379 (0.2379)  loss_classifier: 0.0365 (0.0365)  loss_box_reg: 0.0507 (0.0507)  loss_mask: 0.1434 (0.1434)  loss_objectness: 0.0042 (0.0042)  loss_rpn_box_reg: 0.0030 (0.0030)  time: 0.7387  data: 0.2417  max mem: 2413\n",
      "Epoch: [6]  [5/6]  eta: 0:00:00  lr: 0.000050  loss: 0.2379 (0.2676)  loss_classifier: 0.0348 (0.0371)  loss_box_reg: 0.0507 (0.0642)  loss_mask: 0.1434 (0.1561)  loss_objectness: 0.0042 (0.0048)  loss_rpn_box_reg: 0.0037 (0.0054)  time: 0.5604  data: 0.0419  max mem: 2413\n",
      "Epoch: [6] Total time: 0:00:03 (0.5691 s / it)\n",
      "creating index...\n",
      "index created!\n",
      "Test:  [0/4]  eta: 0:00:01  model_time: 0.1324 (0.1324)  evaluator_time: 0.0152 (0.0152)  time: 0.2957  data: 0.1470  max mem: 2413\n",
      "Test:  [3/4]  eta: 0:00:00  model_time: 0.1223 (0.1278)  evaluator_time: 0.0123 (0.0139)  time: 0.1808  data: 0.0379  max mem: 2413\n",
      "Test: Total time: 0:00:00 (0.1898 s / it)\n",
      "Averaged stats: model_time: 0.1223 (0.1278)  evaluator_time: 0.0123 (0.0139)\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "IoU metric: bbox\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.340\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.862\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.086\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.350\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.320\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.375\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.500\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.500\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.350\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.650\n",
      "IoU metric: segm\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.563\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.862\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.862\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.650\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.502\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.475\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.675\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.675\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.700\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.650\n",
      "Epoch: [7]  [0/6]  eta: 0:00:04  lr: 0.000050  loss: 0.2438 (0.2438)  loss_classifier: 0.0305 (0.0305)  loss_box_reg: 0.0671 (0.0671)  loss_mask: 0.1339 (0.1339)  loss_objectness: 0.0038 (0.0038)  loss_rpn_box_reg: 0.0085 (0.0085)  time: 0.7469  data: 0.2429  max mem: 2413\n",
      "Epoch: [7]  [5/6]  eta: 0:00:00  lr: 0.000050  loss: 0.2460 (0.2666)  loss_classifier: 0.0336 (0.0372)  loss_box_reg: 0.0643 (0.0636)  loss_mask: 0.1436 (0.1562)  loss_objectness: 0.0036 (0.0042)  loss_rpn_box_reg: 0.0048 (0.0054)  time: 0.5618  data: 0.0420  max mem: 2413\n",
      "Epoch: [7] Total time: 0:00:03 (0.5672 s / it)\n",
      "creating index...\n",
      "index created!\n",
      "Test:  [0/4]  eta: 0:00:01  model_time: 0.1346 (0.1346)  evaluator_time: 0.0164 (0.0164)  time: 0.3027  data: 0.1506  max mem: 2413\n",
      "Test:  [3/4]  eta: 0:00:00  model_time: 0.1223 (0.1268)  evaluator_time: 0.0132 (0.0142)  time: 0.1803  data: 0.0384  max mem: 2413\n",
      "Test: Total time: 0:00:00 (0.1913 s / it)\n",
      "Averaged stats: model_time: 0.1223 (0.1268)  evaluator_time: 0.0132 (0.0142)\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "IoU metric: bbox\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.340\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.862\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.086\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.350\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.320\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.375\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.500\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.500\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.350\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.650\n",
      "IoU metric: segm\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.574\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.862\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.862\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.667\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.502\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.500\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.675\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.675\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.700\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.650\n",
      "Epoch: [8]  [0/6]  eta: 0:00:04  lr: 0.000050  loss: 0.2010 (0.2010)  loss_classifier: 0.0278 (0.0278)  loss_box_reg: 0.0326 (0.0326)  loss_mask: 0.1362 (0.1362)  loss_objectness: 0.0010 (0.0010)  loss_rpn_box_reg: 0.0033 (0.0033)  time: 0.7428  data: 0.2383  max mem: 2413\n",
      "Epoch: [8]  [5/6]  eta: 0:00:00  lr: 0.000050  loss: 0.2525 (0.2622)  loss_classifier: 0.0340 (0.0365)  loss_box_reg: 0.0653 (0.0631)  loss_mask: 0.1378 (0.1541)  loss_objectness: 0.0035 (0.0032)  loss_rpn_box_reg: 0.0047 (0.0053)  time: 0.5429  data: 0.0424  max mem: 2413\n",
      "Epoch: [8] Total time: 0:00:03 (0.5489 s / it)\n",
      "creating index...\n",
      "index created!\n",
      "Test:  [0/4]  eta: 0:00:01  model_time: 0.1322 (0.1322)  evaluator_time: 0.0160 (0.0160)  time: 0.2997  data: 0.1504  max mem: 2413\n",
      "Test:  [3/4]  eta: 0:00:00  model_time: 0.1227 (0.1286)  evaluator_time: 0.0134 (0.0142)  time: 0.1816  data: 0.0379  max mem: 2413\n",
      "Test: Total time: 0:00:00 (0.1887 s / it)\n",
      "Averaged stats: model_time: 0.1227 (0.1286)  evaluator_time: 0.0134 (0.0142)\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "IoU metric: bbox\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.346\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.876\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.086\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.350\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.334\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.375\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.500\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.500\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.350\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.650\n",
      "IoU metric: segm\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.582\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.876\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.876\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.667\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.551\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.500\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.675\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.675\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.700\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.650\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: [9]  [0/6]  eta: 0:00:04  lr: 0.000005  loss: 0.2393 (0.2393)  loss_classifier: 0.0313 (0.0313)  loss_box_reg: 0.0656 (0.0656)  loss_mask: 0.1335 (0.1335)  loss_objectness: 0.0023 (0.0023)  loss_rpn_box_reg: 0.0066 (0.0066)  time: 0.7595  data: 0.2422  max mem: 2413\n",
      "Epoch: [9]  [5/6]  eta: 0:00:00  lr: 0.000005  loss: 0.2374 (0.2618)  loss_classifier: 0.0321 (0.0366)  loss_box_reg: 0.0527 (0.0630)  loss_mask: 0.1365 (0.1540)  loss_objectness: 0.0025 (0.0029)  loss_rpn_box_reg: 0.0047 (0.0053)  time: 0.5471  data: 0.0426  max mem: 2413\n",
      "Epoch: [9] Total time: 0:00:03 (0.5524 s / it)\n",
      "creating index...\n",
      "index created!\n",
      "Test:  [0/4]  eta: 0:00:01  model_time: 0.1325 (0.1325)  evaluator_time: 0.0160 (0.0160)  time: 0.3006  data: 0.1510  max mem: 2413\n",
      "Test:  [3/4]  eta: 0:00:00  model_time: 0.1224 (0.1269)  evaluator_time: 0.0132 (0.0141)  time: 0.1800  data: 0.0381  max mem: 2413\n",
      "Test: Total time: 0:00:00 (0.1871 s / it)\n",
      "Averaged stats: model_time: 0.1224 (0.1269)  evaluator_time: 0.0132 (0.0141)\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "Accumulating evaluation results...\n",
      "DONE (t=0.00s).\n",
      "IoU metric: bbox\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.349\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.876\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.131\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.350\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.339\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.375\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.575\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.575\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.350\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.800\n",
      "IoU metric: segm\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.584\n",
      " Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.876\n",
      " Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.842\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.667\n",
      " Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.535\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.500\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.700\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.700\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.700\n",
      " Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.700\n"
     ]
    }
   ],
   "source": [
    "# let's train it for 10 epochs\n",
    "num_epochs = 10\n",
    "\n",
    "for epoch in range(num_epochs):\n",
    "    # train for one epoch, printing every 10 iterations\n",
    "    train_one_epoch(model, optimizer, data_loader, device, epoch, print_freq=10)\n",
    "    # update the learning rate\n",
    "    lr_scheduler.step()\n",
    "    # evaluate on the test dataset\n",
    "    evaluate(model, data_loader_test, device=device)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# pick one image from the test set\n",
    "img, _ = dataset_test[1]\n",
    "# put model into evaluation mode\n",
    "model.eval()\n",
    "with torch.no_grad():\n",
    "    prediction = model([img.to(device)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Let's inspect the image and the predicted segmentation masks.\n",
    "Image.fromarray(img.mul(255).permute(1, 2, 0).byte().numpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "Image.fromarray(prediction[0]['masks'][0, 0].mul(255).byte().cpu().numpy())"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}